Design and Performance of a Hydroponic Bioreactor for Removing Emerging Contaminants from Wastewater Effluent

Abstract

Emerging contaminants have been identified in detectable quantities in wastewater effluents around the world. These emerging contaminants include pharmaceuticals, personal care products, industrial and household chemicals, as well as endocrine disrupting compounds that are not removed through traditional wastewater treatment processes. A small-scale novel bioreactor was designed and optimized to remove these contaminants from tertiary treated wastewater effluent from a water reclamation plant in Tucson, Arizona. More specifically, this bioreactor was designed to treat 15 liters of effluent by circulating it through a hydroponic media bed bioreactor. Four hydroponic media— 25.4-mm lava rock, 12.7-mm (small) fractured rock, 25.4-mm (large) fractured rock and 12.0-mm expanded clay pebbles (LECA)—underwent hydrodynamic tests to determine which provided optimal flow conditions for maximizing dispersion and mixing within the bioreactor. Tracer tests were performed for each media with a sodium chloride tracer and used to determine the mean residence times (MRT) and vessel dispersion numbers at two different flow rates (6.9 and 9.7 L min-1) and two water levels (11.8-12.8 and 20-20.8 cm). In the low water level treatments, expanded clay had a 16-29% longer MRT (148.7±8.0 s) than the other media (small rock - 128.1±17.3 s; large rock - 114.±14.1 s; lava rock -121.0±1.4 s) in the low flow condition and a significantly longer MRT (115.7±8.3 s) than the small fractured rock ( 89.7±0.6 s) and the lava rock (95.0±7.1 s) in the high flow condition but not the large fractured rock (108.0±21.5 s). In the high-water level treatments, the small fractured rock had a significantly longer MRT (203.1±13.4 s) than the other media (LECA - 135.7±9.9 s; Large rock - 137.8±12.4 s) in the low flow condition but only a longer MRT (138.7±2.3 s) than the expanded clay pebbles (118.9±13.8 s) in the high flow condition, not the large fractured rock (152.3±16.3 s). The estimated axial water velocity at the high-water level with low flow was the slowest of all treatments tested (0.32 – 0.38 cm s-1), but interestingly, only the small fractured rock’s MRT was significantly correlated to the estimated water velocity. While the MRT of the bioreactor with small fractured rock decreased as the velocity increased, the MRTs in the expanded clay treatments did not show any significant differences between flow rates or water volumes. The vessel dispersion number (low flow - 14.0±2.5 and high flow - 15.3±10.6), and hence the dispersion through the LECA was greater than the other media tested (small rock – 2.6±0.6 and 4.2±0.3; large rock - 4.3±0.4 and 8.9±2.3; lava rock - 6.6±2.2 and 5.2±0.8) at the low water level, but statistical significance was only achieved under the low flow condition. Vessel dispersion numbers for bioreactors with the LECA and large fractured rock were not affected by flow rate or water volume. However, dispersion in the small fractured rock treatment was significantly greater at the lowest estimated water velocity tested (0.38 cm s-1) than the other three (0.54 cm s-1, 0.61 cm s-1, 0.86 cm s-1). Tracers were also used to determine the mixing times within bioreactors with the four media. Mixing time (low flow – 267±157 s and high flow 318±2 s) was significantly shorter with LECA in the low water level treatments (small rock – 336±17 s and 782±147 s; large rock – 854±227 s and 809±175 s; lava rock – 1835±617 s and 390±21 s). In the mixing tests, small fractured rock was comparable to the LECA at the low volume, low flow treatments but mixed 59% slower than in bioreactors with LECA at the higher flow rate. Overall, LECA performed better than the other substrate media tested, while the small fractured rock only performed well at the lowest estimated water velocities (which occurred at the high-water level). The large fractured rock was substandard to both of those media and the lava rock was inconclusive for mixing tests due to the tracer getting caught up in its convoluted channels as it passed through the bioreactors. In the second experiment, bioreactors filled with LECA were planted with high and low densities of sorghum (Sorghum bicolor), switchgrass (Panicum virgatum) and Bacillus thuringiensis (BT) cotton (Gossypium sp.) in addition to three unplanted bioreactors. After an acclimation period with hydroponic nutrients, the bioreactors were drained and rinsed, then filled with tertiary wastewater effluent. Water samples were taken at time zero and then after one, two, five and 10 days and analyzed using LC-MS/MS to determine the concentrations of emerging contaminants present. Sixteen emerging contaminants were found at detectable quantities in the effluent—Atenolol, Benzotriazole, Carbamazepine, DEET, Diclofenac, Diphenhydramine, Hydrochlorothiazide, Ibuprofen, Iohexol, Iopamidol, Iopromide, Primidone, Simazine, Sucralose, Sulfamethoxazole and TCPP. After one day, both the planted and unplanted bioreactors were able to remove at least 97% and 89% (below detection limits) of ibuprofen and diphenhydramine, respectively compared to a control with no media or plants that had zero removal. Atenolol was shown to be significantly removed in all bioreactors after two days and removed below detection limits after five days compared to only 50% removal in the control. After five days, both planted and unplanted bioreactors were able to remove nearly 80% or more of benzotriazole, carbamazepine, hydrochlorothiazide, Iohexol, Iopamidol, Iopromide, Primidone, Simazine, Sulfamethoxazole and TCPP compared to nearly zero percent removal in the control. After 10 days, DEET was removed below detection limits and nearly 60% of sucralose was removed. Also, after 10 days, diclofenac, iopromide, primidone, simazine were all removed below detection limits with still near zero percent removal in the control. The planted bioreactors showed significantly more removal of benzotriazole than the unplanted bioreactors after five and 10 days by roughly 15%. Bioreactors planted with cotton had significantly more removal of sulfamethoxazole than unplanted bioreactors by 16-19% after five days and 18-20% more removal after 10 days. Overall, the bioreactors were able to successfully reduce levels of all emerging contaminants found in tertiary wastewater effluent. Based on the experiment, the plants in two cases served as a major mechanism for removal of emerging contaminants. For the other 14 cases, where plants did not serve as the major removal mechanism, the most likely mechanism involved was sorption to media, degradation by the microbiome within the bioreactors, or both. The hydrodynamics were improved within the bioreactors, two baffle configurations were then examined using tracer tests at two flow rates. A split baffle configuration proved to have a longer mean residence time (176.9±14.8 s) and significantly greater vessel dispersion number (11.6±3.4) than a serpentine configuration (134.0±4.1 s and 6.4±0.3) and a no baffle configuration (122.9±0.7 s and 5.6±0.1) at the low flow level. At the high flow rate however, the mean residence times were not significantly different between any of the configurations (split – 105.5±0.4 s; serpentine – 100.5±2.1 s; no-baffle 114.78.0± s) but there was significantly more dispersion in the no-baffle configuration (12.8±2.8) compared to the split baffle (3.9±0.1) and serpentine baffle (4.9±0.2) configurations. Therefore, it is concluded that a hydroponic bioreactor with a split baffle configuration, filled with LECA will maximize the removal of emerging contaminants when used to treat tertiary-treated wastewater effluent if the estimated water velocity is not too high. Although cotton appeared to outperform the other two plants tested, more research needs to be done focusing on the removal by different plants. Future research should focus on the specific pathways of contaminant removal within a hydroponic media bed bioreactor and how those pathways can be optimized for maximum removal of emerging contaminants.